Convergent organization of aberrant MYB complex controls oncogenic gene expression in acute myeloid leukemia - eLife
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
RESEARCH ARTICLE
Convergent organization of aberrant MYB
complex controls oncogenic gene
expression in acute myeloid leukemia
Sumiko Takao1,2†, Lauren Forbes1,2,3†, Masahiro Uni1,2†, Shuyuan Cheng1,2,3,
Jose Mario Bello Pineda4,5,6,7, Yusuke Tarumoto8,9, Paolo Cifani1,
Gerard Minuesa1, Celine Chen1, Michael G Kharas1,3, Robert K Bradley4,6,7,
Christopher R Vakoc8, Richard P Koche10, Alex Kentsis1,2,3*
1
Molecular Pharmacology Program, Sloan Kettering Institute, New York, United
States; 2Tow Center for Developmental Oncology, Department of Pediatrics,
Memorial Sloan Kettering Cancer Center, New York, United States; 3Departments
of Pharmacology and Physiology & Biophysics, Weill Cornell Graduate School of
Medical Sciences, Cornell University, New York, United States; 4Computational
Biology Program, Public Health Sciences Division, Fred Hutchinson Cancer Research
Center, Seattle, United States; 5Medical Scientist Training Program, University of
Washington, Seattle, United States; 6Basic Sciences Division, Fred Hutchinson
Cancer Research Center, Seattle, United States; 7Department of Genome Sciences,
University of Washington, Seattle, United States; 8Cold Spring Harbor Laboratory,
Cold Spring Harbor, United States; 9Institute for Frontier Life and Medical Sciences,
Kyoto University, Kyoto, Japan; 10Center for Epigenetics Research, Sloan Kettering
Institute, New York, United States
*For correspondence: Abstract Dysregulated gene expression contributes to most prevalent features in human
kentsisresearchgroup@gmail.com cancers. Here, we show that most subtypes of acute myeloid leukemia (AML) depend on the
†
aberrant assembly of MYB transcriptional co-activator complex. By rapid and selective
These authors contributed
peptidomimetic interference with the binding of CBP/P300 to MYB, but not CREB or MLL1, we find
equally to this work
that the leukemic functions of MYB are mediated by CBP/P300 co-activation of a distinct set of
Competing interest: See transcription factor complexes. These MYB complexes assemble aberrantly with LYL1, E2A, C/EBP
page 34 family members, LMO2, and SATB1. They are organized convergently in genetically diverse
Funding: See page 34 subtypes of AML and are at least in part associated with inappropriate transcription factor co-
expression. Peptidomimetic remodeling of oncogenic MYB complexes is accompanied by specific
Received: 18 December 2020
proteolysis and dynamic redistribution of CBP/P300 with alternative transcription factors such as
Accepted: 01 February 2021
Published: 02 February 2021 RUNX1 to induce myeloid differentiation and apoptosis. Thus, aberrant assembly and sequestration
of MYB:CBP/P300 complexes provide a unifying mechanism of oncogenic gene expression in AML.
Reviewing editor: Irwin This work establishes a compelling strategy for their pharmacologic reprogramming and
Davidson, Institut de Génétique therapeutic targeting for diverse leukemias and possibly other human cancers caused by
et de Biologie Moléculaire et
dysregulated gene control.
Cellulaire, France
Copyright Takao et al. This
article is distributed under the
terms of the Creative Commons
Attribution License, which
Introduction
permits unrestricted use and Gene dysregulation is one of the most prevalent features in human cancers (Bradner et al., 2017).
redistribution provided that the In many tumors, this is due to the pathogenic mutations of promoters, enhancers, and genes encod-
original author and source are ing either transcription factors or factors that regulate chromatin and gene expression. In blood can-
credited. cers, and acute myeloid leukemias (AML) in particular, aberrant gene expression is thought to
Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 1 of 46
Research article Cancer Biology
contribute to most important properties of leukemia cells, including self-renewal, growth, and resis-
tance to therapy. For example, numerous pathogenic chromosomal translocations in AML, such as
those involving AML1 (RUNX1) and MLL1 (KMT2A) produce chimeric transcription or chromatin
remodeling factors that cause disease (Look, 1997). Consequently, therapeutic strategies aimed at
restoring normal gene expression are compelling because of their ability to target the causal molec-
ular processes and induce leukemia cell differentiation and elimination, leading in principle to dura-
ble disease control.
While specific molecular dependencies have been identified for some genetic subtypes of AML,
such as DOT1L or Menin inhibition for MLL-rearranged leukemias (Krivtsov et al., 2019), and
CARM1 inhibition for AML1-rearranged leukemias (Greenblatt et al., 2019), distinct pathogenetic
mechanisms of diverse AML subtypes also appear to converge on shared molecular pathways. For
example, approximately 25% of adult and childhood AMLs, including both MLL-rearranged and non-
rearranged cases, require aberrant activation of the transcription factor MEF2C, conferring suscepti-
bility to MARK and SIK inhibitors, which are currently being explored for clinical trials for patients
(Brown et al., 2018; Tarumoto et al., 2018; Vakoc and Kentsis, 2018). Similarly, nearly 50% of
examined AML specimens exhibit aberrant activation of HGF/MET/FGFR signaling (Kentsis et al.,
2012) and are being currently targeted therapeutically in the ongoing clinical trial of combined MET
and FGFR inhibitors in patients with relapsed or refractory AML (ClinicalTrials.gov Identifier
NCT03125239). Even for therapies targeting leukemogenic proteins directly, such as inhibitors of
IDH1/2, FLT3, KIT, SYK, as well as epigenetic and apoptotic therapies such as decitabine and vene-
toclax, their therapeutic efficacy and resistance depend on the underlying gene expression pheno-
typic states of AML cells (Tyner et al., 2018). Thus, there is intense interest in defining shared
molecular dependencies controlling leukemogenic gene expression in AML that can provide effec-
tive therapeutic options for patients.
Recently, MYB has emerged as a therapeutic target in AML, as transient suppression of Myb
nearly completely eliminates leukemia development in mouse models in vivo while sparing normal
hematopoietic cells (Zuber et al., 2011). Indeed, pioneering studies have implicated Myb as a key
mediator of leukemias (Klempnauer and Bishop, 1984; Luger et al., 2002). MYB is the cellular
homologue of the viral v-Myb oncogene that can cause avian leukemias and function as a pioneer
transcription factor in mammalian cells (Biedenkapp et al., 1988). MYB functions as a master regula-
tor of gene expression in diverse cell types, including hematopoietic cells where it controls cell pro-
liferation and differentiation (Ramsay and Gonda, 2008). Both mutations and translocations of MYB
have causal roles in various human malignancies, including leukemias. For example, aberrant expres-
sion of TAL1 in T-cell acute lymphoid leukemia (T-ALL) is induced by pathogenic somatic mutations
that create neomorphic MYB-binding sites (Mansour et al., 2014). Likewise, MYB is recurrently rear-
ranged in distinct subtype of blastic plasmacytoid dendritic cell neoplasms (BPDCN), a highly refrac-
tory hematologic malignancy (Suzuki et al., 2017).
Notably, the Booreana strain of mice that impairs the binding of Myb by its co-activator CBP/
P300 (Crebbp/Ep300) due to the mutation of Myb E308G in its transcriptional activation domain is
resistant to leukemogenesis induced by the otherwise fully penetrant MLL-AF9 and AML1-ETO onco-
genes but has largely normal hematopoiesis (Pattabiraman et al., 2014). Altogether, these findings
indicate that MYB and its co-factor CBP/P300 are fundamentally dysregulated in AML, presumably
through disordered gene expression that characterizes most forms of this disease. However, the
specific details of this mechanism remain poorly understood, largely due to the lack of suitable
tools.
Recently, we developed a peptidomimetic inhibitor of MYB:CBP/P300 (Ramaswamy et al.,
2018). Here, we report its second-generation version that has significantly increased potency, and
consequently suppresses leukemic MYB functions in most AML subtypes tested, while relatively spar-
ing normal hematopoietic progenitor cells. By rapid and selective peptidomimetic interference with
the binding of CBP/P300 to MYB, but not CREB or MLL1, we find that the leukemic functions of
MYB are mediated by CBP/P300-mediated co-activation of a distinct set of transcriptional factor
complexes that are aberrantly assembled with MYB in AML cells, which is associated at least in part
with their inappropriate expression. This therapeutic remodeling is accompanied by dynamic redistri-
bution of CBP/P300 complexes to genes that control cellular differentiation and growth. These find-
ings provide a unifying mechanism of oncogenic gene control, involving aberrant assembly of
transcription factor complexes and sequestration of CBP/P300 to promote oncogenic gene
Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 2 of 46
Research article Cancer Biology
expression and block cellular differentiation. This paradigm should apply to other human cancers
caused by dysregulated gene control, elucidate specific molecular determinants of leukemia patho-
genesis, and enable the development of definitive therapies for patients.
Results
Genome-wide CRISPR screen identifies CBP requirement for the
susceptibility of AML cells to peptidomimetic blockade of MYB:CBP/
P300
In prior work, we used genetic and structural evidence to design a peptidomimetic inhibitor of the
MYB:CBP/P300 transcription coactivation complex, termed MYBMIM (Ramaswamy et al., 2018). To
elucidate its mechanisms of action in an unbiased manner, we carried out a genome-wide CRISPR
knockout screen to identify genes whose depletion affects the susceptibility of AML cells to MYB-
MIM. We used MOLM13 cells stably expressing Cas9 and transduced them with lentiviruses encod-
ing the genome-wide GeCKOv2 library at a multiplicity of infection of 0.3 (Figure 1A). Following
selection of transduced cells, we treated them with MYBMIM or PBS control in independent biologi-
cal replicates, and quantified the enrichment and depletion of cell clones expressing specific sgRNAs
by DNA sequencing of lentiviral barcodes (Figure 1A). This screen revealed a variety of genes whose
depletion confers relative resistance and susceptibility to MYBMIM treatment, consistent with the
presence of diverse cellular pathways that regulate oncogenic gene expression (Supplementary file
2a). The most significantly affected gene whose depletion was required to confer resistance to MYB-
MIM was CBP (Figure 1B). In contrast, CBP depletion exhibited no enrichment upon PBS treatment
(Figure 1—figure supplement 1). Thus, MYBMIM is a specific inhibitor of MYB:CBP, and emphasizes
the exquisite specificity of peptidomimetic inhibitors as pharmacologic modulators of protein
interactions.
CRYBMIM is a peptidomimetic chimera that specifically binds CBP/P300
KIX domain
Of all the functional genetic dependencies examined to date, the transcription factor MYB demon-
strates the broadest dependency across diverse AML subtypes, as compared to other non-hemato-
poietic cancers (Tarumoto et al., 2018). To generalize this analysis, we queried 688 human cancer
cell lines tested as part of the DepMap Cancer Dependency Map to identify genes that are selec-
tively required for the growth and survival of leukemia as compared to other cancer types. We found
that MYB is the most significantly required human gene in 37 leukemia cell lines, including 20 AML
cell lines, of diverse molecular subtypes (p=1.1e-15; Figure 2A).
MYB target gene activation requires its specific interaction with CREB-binding protein (CBP)/
P300 for co-activation (Dai et al., 1996). The helical MYB transactivation domain comprising residues
293–310 binds to the KIX domain of CBP/P300 (Zor et al., 2004). Using molecular mechanics simula-
tions, we previously developed a peptidomimetic inhibitor of this interaction, termed MYBMIM
(Ramaswamy et al., 2018). MYBMIM uses stereoselective substitution of D-amino acids to confer
proteolytic stability, and the cationic TAT domain for cell penetration. As a result, MYBMIM can spe-
cifically inhibit MYB:CBP/P300 binding in cells, but its activity is less pronounced in non-MLL-rear-
ranged AML cells (Ramaswamy et al., 2018). Given that MYBMIM bound to recombinant CBP KIX
domain with the dissociation constant of 21.3 ± 2.9 mM, as compared to the native MYB peptide of
4.2 ± 0.5 mM (Ramaswamy et al., 2018), we reasoned that a peptidomimetic inhibitor with higher
affinity to CBP/P300 would be more effective.
Consequently, we used molecular modeling to extend MYBMIM into the adjoining binding site
that binds CREB (Radhakrishnan et al., 1997; Cheng et al., 2008). We appended CREB residues
124–147 to MYBMIM, while replacing the EKIRK motif to maintain favorable backbone geometry, as
confirmed by molecular energy minimization calculations in implicit solvent (Figure 2B and C and
Figure 2—figure supplement 1). Termed CRYBMIM, this design preserves key MYB residues impli-
cated in leukemic transformation, including E308 which forms a salt bridge with CBP, while including
the pS133-containing portion of CREB that is responsible for its high-affinity binding to CBP
(Zor et al., 2004; Cheng et al., 2008; Radhakrishnan et al., 1997). Similarly, we also designed two
additional peptidomimetic inhibitors targeting the distinct CREB and MLL1-binding sites that are
Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 3 of 46
Research article Cancer Biology
Figure 1. CBP depletion is required to confer resistance to MYBMIM in AML cells. (A) Schematic of the genome-wide CRISPR screen to identify genes
whose loss modifies MYBMIM effects in MOLM13 cells expressing Cas9. Cells were transduced with the GeCKOv2 library expressing single sgRNAs at
low multiplicity of infection, followed by 3-day treatment with 10 mM MYBMIM versus PBS control, with the sgRNA representation assessed by DNA
sequencing. (B) Volcano plot showing the relative abundance of cell clones expressing sgRNAs targeting specific genes (fold change of MYBMIM-
treated cells at T1 versus T0) and their statistical significance from biological replicates. Dashed line represents no enrichment, with positive values
representing genes whose depletion confers relative resistance to MYBMIM. CBP is marked in red.
The online version of this article includes the following source data and figure supplement(s) for figure 1:
Source data 1. Genome-wide CRISPR knockout screen (GeCKO) gene summary.
Figure supplement 1. CBP depletion is dispensable in AML cells.
proximal to but non-overlapping with the MYB-binding site, termed CREBMIM and MLLMIM,
respectively (Figure 2B and C and Figure 2—figure supplement 1A–C).
As expected, both CRYBMIM and CREBMIM bind to the purified recombinant CBP KIX domain
with significantly improved (three– to six-fold) affinities as compared to MYBMIM and measured by
microscale thermophoresis (Kd of 5.7 ± 0.2 mM, 2.9 ± 0.7 mM, and 17.3 ± 1.6 mM, p=1e-15;
Figure 2D). To confirm these peptides can bind the CBP/P300 complex from cells, we immobilized
biotinylated CRYBMIM and CREBMIM peptides on streptavidin beads and used them to affinity
purify CBP/P300 from non-denatured nuclear extracts of MV411 AML cells. Consistently, we
observed efficient binding of nuclear CBP/P300 to peptide-conjugated but not control streptavidin
Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 4 of 46
Research article Cancer Biology Figure 2. CRYBMIM is an improved peptidomimetic chimera that specifically binds the KIX domain of CBP/P300 in vitro and in cells. (A) Heatmap of the top 10 and bottom 10 gene dependencies for survival and proliferation of 652 cancer cell lines in the DepMap Cancer Dependency Map Project, ranked by the greatest dependency for 37 leukemia lines, 20 of which are AML cell lines, as indicated by the red color gradient; *p=1.1e-15, t-test of leukemia versus other tumor types. (B) Retro-inverso amino acid sequences of MYBMIM, CREBMIM and CRYBMIM, with amino acids derived from MYB, Figure 2 continued on next page Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 5 of 46
Research article Cancer Biology
Figure 2 continued
CREB, and TAT marked in blue, green, and black, respectively. (C) Molecular model of the CRYBMIM:KIX complex. Residues making contact with KIX
are labeled, with portions derived from MYB and CREB marked in blue and green, respectively. (D) Binding of FITC-conjugated CREBMIM (blue),
CRYBMIM (red), and MYBMIM (black), as measured using microscale thermophoresis; Kd = 2.9 ± 0.7, 5.7 ± 0.2, and 17.3 ± 1.6, respectively. Error bars
represent standard deviations of three biological replicates; p100 mM, Figure 2E–F), consistent with the much higher nM affinity and allo-
steric effects that characterize the CREB:CBP/P300 interaction (Goto et al., 2002;
Radhakrishnan et al., 1997). This is also consistent with the presence of distinct CBP transcription
factor complexes in AML cells, some of which (MYB) are susceptible to peptidomimetic blockade,
and others (CREB) exhibiting more stable interactions. Importantly, CRYBMIM binds CBP/P300 spe-
cifically, as exposure of AML cell extracts to streptavidin-immobilized biotinylated CRYBMIM leads
to efficient binding to CBP/P300, but not MED15 which is highly expressed in MV411 AML cells and
contains a known KIX domain with the closest sequence similarity to CBP/P300 (38% identity; Fig-
ure 2—figure supplement 1D–E). Thus, CRYBMIM is a specific high-affinity inhibitor of MYB:CBP/
P300 binding.
Potent and broad-spectrum activity of CRYBMIM against diverse
subtypes of AML
To confirm that CRYBMIM maintains effective cell penetration and nuclear accumulation in AML
cells, we studied its fluorescein isothiocyanate (FITC)-conjugated derivative using live cell confocal
fluorescence microscopy (Figure 3A). Consistently, we observed that both FITC-CRYBMIM and
FITC-CREBMIM efficiently localized to the nuclei of MV411 AML cells within 1 hr of peptide treat-
ment (Figure 3A). We confirmed specific nuclear accumulation of CRYBMIM as opposed to non-spe-
cific membrane binding that can affect TAT-containing peptides by confocal subcellular imaging of
cells co-stained with specific mitochondrial and nuclear dyes (Figure 3A).
Previously, we observed that MYBMIM blocks the binding of MYB to CBP/P300 in AML cells,
requiring relatively high 20 mM concentrations for 3 hr to achieve this effect (Ramaswamy et al.,
2018). To ascertain whether CRYBMIM can achieve more potent interference with the binding of
MYB to the CBP/P300 complex in cells due to its improved affinity as compared
to MYBMIM (Figure 2D), we treated MV411 cells with 10 mM peptides for 1 hr and immunoprecipi-
tated CBP/P300 using specific antibodies (Figure 3B). Under these more stringent conditions, we
found that CRYBMIM is indeed more potent compared to MYBMIM, as evidenced by the substantial
depletion of MYB from the immunoprecipitated CBP/P300 complex by CRYBMIM but not MYBMIM
under these conditions (Figure 3B).
This effect was specific because the inactive analogue of CRYBMIM, termed CG3, in which three
key residues have been replaced with glycines (Supplementary file 1a), was unable to compete with
MYB:CBP/P300 binding in cells, an effect observed with CBP/P300-specific but not control isotype
non-specific antibodies (Figure 3B). Neither CREBMIM nor CRYBMIM treatment interfered with the
binding of CREB to the cellular CBP/P300 complex (Figure 3B), in agreement with the affinities of
their direct binding to the recombinant CBP KIX domain (Figure 2D–F), the nM affinity of CREB:
CBP/P300 interaction (Radhakrishnan et al., 1997), and the specific molecular features required for
MYB but not CREB or MLL1 binding (Figure 2C and Figure 2—figure supplement 1).
Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 6 of 46
Research article Cancer Biology Figure 3. Potent and broad-spectrum activity of CRYBMIM against diverse AML cell lines but relatively sparing of normal hematopoietic progenitor cells. (A) Representative live cell confocal microscopy images of MV411 cells treated with 100 nM FITC-conjugated peptides as indicated for 1 hr, counterstained with Mitotracker (red) and Hoechst 33342 (blue). Scale bar indicates 20 mm, with z-stack of 1.5 mm. (B) Western blots showing immunoprecipitated nuclear CBP/P300 co-purified with MYB and CREB from MV411 cells treated with 10 mM peptides as indicated for 1 hr. (C–D) Cell viability of MV411 cells as a function of increasing concentrations of 48 hr peptide treatment, comparing (C) CRYBMIM to MYBMIM, CREBMIM and MLLMIM (IC50 = 6.88 ± 3.39, 13.1 ± 3.3 29.15 ± 3.79, and 24.22 ± 2.00, p
Research article Cancer Biology
Figure 3 continued
umbilical cord blood (CD34+ CB, gray) and MV411 AML (red) cells following CRYBMIM or MYBMIM treatment. Data represent three biological
replicates; *p=7.4e-5, t-test of normal CD34+ CB versus MV411 AML cells. (F) Preservation of clonogenic capacity of CD34+ CB cells in differentiating
into erythroid blast forming units (BFU-E, light gray) and granulocyte macrophage colony forming units (CFU-GM, gray) as a function of CRYBMIM
treatment. (G) Cell viability of AML cell lines treated with control PBS (black), 20 mM CREBMIM (blue), MYBMIM (gray), or CRYBMIM (red) as indicated
for 6 days with media replacement every 48 hr in three biological replicates (p=8.6e-3, t-test for CRYBMIM versus control).
The online version of this article includes the following figure supplement(s) for figure 3:
Figure supplement 1. CRYBMIM relatively spares normal hematopoietic progenitor cells in vitro.
Figure supplement 2. Activity of CRYBMIM on non-hematopoietic cells.
To test the prediction that higher affinity binding of CRYBMIM to CBP/P300 would translate into
improved anti-leukemia potency, we assessed its effects on the viability of cultured human leukemia
and normal hematopoietic cells. Consistent with this prediction, we observed that CRYBMIM exhib-
ited significantly higher potency against MV411 AML cells, as compared to MYBMIM as well as
CREBMIM and MLLMIM (IC50 = 6.9 ± 3.4 mM, 13 ± 3.3, 29 ± 3.8 mM, and 24 ± 2.0 mM, p=1e-15;
Figure 3C). We confirmed CRYBMIM’s specificity by analyzing its inactive analogue CG3 and MYB-
MIM’s inactive analogue TG3, in which three residues forming key electrostatic and hydrophobic
interactions with the KIX domain were replaced by glycines; both exhibited significantly reduced
activity (IC50 of 17 ± 1.0 mM and 49 ± 2.6 mM, p=2.75e-4 and
Research article Cancer Biology Figure 4. CBP but not P300 is dispensable for the growth and survival of AML cells, and is required for the susceptibility to peptidomimetic MYB blockade by CRYBMIM. (A) Schematic of the competitive assays to define specific genetic dependencies. AML cells expressing Cas9 and GFP are transduced with sgRNAs targeting specific genes and expressing mCherry, followed by quantitation of cell abundance by fluorescence activated cell scanning (FACS) of GFP and mCherry-expressing cells. (B) Western blots demonstrating specific depletion of CBP in MV411 (left) and MOLM13 (right) Figure 4 continued on next page Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 9 of 46
Research article Cancer Biology
Figure 4 continued
cells expressing sgCBP-1 and sgCBP-2, as compared to control sgNEG1 and sgNEG2 targeting the AAVS1 safe harbor locus. EP300 is shown for
specificity, and Actin serves as the loading control. (C) Relative growth of GFP-expressing MOLM13 cells expressing mCherry and unique sgRNAs
targeting AAVS1 control (sgNEG1 and sgNEG2), CBP (sgCBP-1 and sgCBP-2), and P300 (sgEP300-1 and sgEP300-2) and quantified by FACS on day 0,
3, and 6 after 2-day treatment with 10 mM CRYBMIM or PBS. Data represent biological triplicates of at least 10,000 cells per condition; *p=6.0e-10,
**p=4.1e-8, t-test for day 6 versus day 0 of CRYBMIM treatment of CBP-depleted cells. (D) Analogous experiment as (C) using MV411 cells; *p=1.1e-6,
**p=3.7e-6, t-test for day 6 versus day 0 of CRYBMIM treatment of CBP-depleted cells.
The online version of this article includes the following figure supplement(s) for figure 4:
Figure supplement 1. Genetic dependencies of CRYBMIM susceptibility in AML cells.
harbor locus and CDK1 that is required for cell survival, as negative and positive controls, respec-
tively. We confirmed specific depletion of the majority of CBP protein by Western immunoblotting
in cells expressing sgCBP-1 and sgCBP-2 with intact EP300 expression, but not those not transduced
with sgRNA lentiviruses or those expressing sgNEG-1 and sgNEG-2 (Figure 4B). We found that
depletion of EP300 caused gradual decrease in cell proliferation over 3–6 days (Figure 4—figure
supplement 1A–C), whereas depletion of CDK1 caused acute cell cycle arrest and apoptosis (Fig-
ure 4—figure supplement 1A–C). In contrast, depletion of CBP had no significant effects on steady-
state cell fitness, but caused significant resistance to CRYBMIM, where CBP-deficient cells outcom-
peted their non-transduced counterparts when treated with CRYBMIM (p=6.0e-10 and 4.0e-8 for
sgCBP-1 and sgCBP-2 in MOLM13 cells, respectively, and 1.1e-6 and 3.7e-6 for MV411 cells;
Figure 4C–D). In contrast, K562 cells that are mostly resistant to CRYBMIM exhibited selective fit-
ness upon depletion of EP300 (Figure 4—figure supplement 1D). This suggests that CBP and
EP300 contribute to distinct gene expression programs, presumably as part of specific transcription
factor complexes, as required for their susceptibility to peptidomimetic blockade.
CRYBMIM blocks oncogenic MYB gene expression and restores normal
myeloid cell differentiation
The assembly of MYB with CBP/P300 controls gene expression in part due to its transcriptional co-
activation at specific enhancers and promoters (Kasper et al., 2010; Zhao et al., 2011). Previously,
we found that MYBMIM can suppress MYB:CBP/P300-dependent gene expression, leading to AML
cell apoptosis that required MYB-mediated suppression of BCL2 (Ramaswamy et al., 2018). How-
ever, because MYBMIM’s suppression of gene expression was accompanied mostly by apoptosis,
we were unable to discern the molecular mechanisms that directly dysregulate the activity of the
CBP/P300 transcription factor complex in AML cells.
Given that CRYBMIM has increased affinity for CBP/P300 similar to that of native MYB (5.7 ± 0.2
mM and 4.2 ± 0.5 mM, respectively), we reasoned that its improved activity would now permit
detailed kinetic studies to define the specific gene expression programs that are aberrantly activated
in AML cells. Consistent with this prediction, comparison of the effects of CRYBMIM on gene expres-
sion and consequent apoptosis of MV411 cells revealed that 1 and 4 hr exposures led to significant
changes in gene expression with minimal induction of apoptosis (Figure 5—figure supplement 1).
Thus, we used RNA-sequencing (RNA-seq) to define the changes on gene expression genome-wide
upon 1 and 4 hr exposures to CRYBMIM as compared to PBS control (Figure 5A–B). We found that
after 4-hr duration of treatment, CRYBMIM causes significant downregulation of 2869 genes, includ-
ing known MYB target genes MYC, IKZF1, GATA2, and KIT (Figure 5A–C, Supplementary file 2b-
e). Similar to MYBMIM, CRYBMIM also caused significant upregulation of distinct genes, an effect
that was substantially more pronounced upon 4 hr of treatment (4099 genes; 5A-B,
Supplementary file 2b-e). Interestingly, in addition to the expected suppression of MYB target
genes (Figure 5C), gene set enrichment analysis also revealed significant induction of myeloid and
monocyte differentiation programs (Figure 5D–E and Supplementary file 2f). These effects were
specific because in contrast to CRYBMIM, CREBMIM treatment exhibited minimal changes in gene
expression (Figure 5—figure supplement 1C), consistent with its inability to disrupt the cellular
CBP/P300 complex (Figure 3B). For example, CRYBMIM induced significant increases in the AP-1
family transcription factors FOS and JUN, as well as IL6 and CSF1 that control myeloid differentiation
(Figure 5B; Gonda et al., 1993; Selvakumaran et al., 1992). Nearly 40% of the genes induced by
Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 10 of 46Research article Cancer Biology Figure 5. CRYBMIM blocks oncogenic MYB gene expression and restores normal myeloid cell differentiation. (A–B) Volcano plots of normalized gene expression of MV411 cells upon one (A) and four hour (B) treatment with CRYBMIM, as compared to PBS control, with select genes labeled; p-values denote statistical significance of three biological replicates. (C–E) Gene set enrichment analysis of up- and downregulated gene sets: (C) MYB_Q6, (D) GSE9988_LPS_VS_VEHICLE_TREATED_MONOCYTES_UP, and (E) GERY_CEBP_TARGETS_377. (F) Histogram of Annexin V- or CD11b-stained MV411 Figure 5 continued on next page Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 11 of 46
Research article Cancer Biology Figure 5 continued cell fluorescence, treated with CRYBMIM (red), MYBMIM (gray), or control PBS (black); *p
Research article Cancer Biology Figure 6. CRYBMIM remodels MYB and CBP/P300 chromatin complexes in AML cells. (A–B) Volcano plots of relative MYB chromatin occupancy in MV411 cells changes after 1 hr (A) and 4 hr (B) of 20 mM CRYBMIM treatment compared to PBS control, as analyzed by ChIP-seq. Sequence motifs found in CRYBMIM-induced MYB-depleted (left) and MYB-enriched loci (right) are shown. p-Values denote statistical significance of three technical replicates. (C–D) Volcano plots of CBP/P300 chromatin occupancy changes after 1 hr (C) and 4 hr (D) of 20 mM CRYBMIM treatment as compared to Figure 6 continued on next page Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 13 of 46
Research article Cancer Biology
Figure 6 continued
control, as analyzed by ChIP-seq. Sequence motifs found in CRYBMIM-induced CBP/P300-depleted (left) and CBP/P300-enriched loci (right) are shown.
p-Values denote statistical significance of three technical replicates.
The online version of this article includes the following source data and figure supplement(s) for figure 6:
Source data 1. Differentially occupied chromatin loci measured by MYB and CBP/P300 ChIP-seq in CRYBMIM vs PBS treated MV411 cells (1 hr and 4 hr).
Figure supplement 1. CRYBMIM remodels MYB chromatin occupancies in AML cells.
Because MYB is required for the growth and survival of diverse AML subtypes, we reasoned that
its essential non-redundant co-factors can be identified from the analysis of their functional depen-
dencies, as assessed by genetic CRISPR interference (Tsherniak et al., 2017). We assigned the
CRISPR dependency score of CBP itself in MV411 cells as the threshold to identify functionally non-
redundant MYB:CBP co-factors (Figure 7B). We found that these genes encode factors with diverse
molecular functions, including a group of 59 chromatin-associated proteins (Supplementary file 3b).
By using currently annotated protein-protein interactions (Oughtred et al., 2019), we constructed
their interaction network, based on interactions detected by affinity purifications coupled with either
western immunoblotting or mass spectrometry (Figure 7C). Analysis of their expression in normal
human hematopoietic progenitor as compared to AML cells (Cancer Genome Atlas Research Net-
work et al., 2013), led to the identification of candidate co-factors with apparently aberrant expres-
sion in AML cells (Figure 7C, dark red). Indeed, 10 such factors were selectively required for 37
leukemia but not 615 other non-hematopoietic cancer cells lines (Figure 7D). The physical associa-
tion of MYB, CBFB, ZEB2, C/EBP family members, LYL1, SPI1, RUNX1, LMO2, and GFI1 and their
non-redundant functional dependencies in AML cells (Figure 7A–D) are in agreement with the chro-
matin dynamics involving distinct MYB, C/EBP family members, LYL1, SPI1, and RUNX1 DNA-bind-
ing motifs observed in CRYBMIM-treated cells (Figure 6A–B), associated with the apparent
redistribution and remodeling of their CBP/P300 co-activator complexes (Figure 6C–D). Indeed,
these factors directly associate with the MYB regulatory complex, and their DNA-binding motifs are
enriched at loci affected by CRYBMIM treatment (Figure 6).
To define the composition of the MYB transcription factor complex across biologically and geneti-
cally distinct subtypes of AML, we used co-immunoprecipitation to measure the physical association
of MYB and its cofactors in a panel of AML cell lines, spanning representative cell types with rela-
tively high (MV411, HL60, OCIAML2, OCIAML3) and low (U937, Kasumi-1, K562) susceptibility to
CRYBMIM (Figure 8). We found that in all seven cell lines tested, MYB was physically associated
with LYL1 and E2A transcription factors (Figure 8A). In contrast, LMO2 was physically associated
with MYB in all cell lines except OCIAML2, CEBPA was co-assembled with MYB in OCIAML2 and
U937 cells, and SATB1 was co-assembled with MYB in MV411 and HL60 cells (Figure 8A). These
findings are all in complete agreement with the chromatin dynamics of MYB, as observed using
ChIP-seq (Figure 6). In addition, we corroborated these findings by examining the association of
specific transcription factors with CBP/P300 (Figure 8B). The only exception is SPI1/PU.1, which
does not appear to be physically associated with MYB or CBP/P300 in examined cell lines, suggest-
ing that the apparent SPI1/PU.1 sequence motifs observed in MYB-associated loci are due to other
ETS family factors (Figure 6). In all, MYB assembles a convergently organized transcription factor
complex in genetically diverse AML cells.
MYB transcription complexes are associated with aberrant cofactor
expression and assembly in AML cells
Insofar as various forms of AML exhibit blockade of normal hematopoietic differentiation induced by
distinct leukemia oncogenes, and at least some of the MYB-assembled cofactors have reduced gene
expression in normal CD34+ hematopoietic progenitor cells (Figure 7C), we reasoned that their
assembly in leukemia cells may be due to their aberrant co-expression. To test this, we measured
their protein expression in human AML cells using quantitative fluorescent immunoblotting, as com-
pared to normal human CD34+ umbilical cord blood progenitor, adult peripheral blood B- and
T-lymphocytes, and monocytes (Figure 9A). We found that most transcription factors that are
assembled with MYB in diverse AML cell lines could be detected in one or more normal human
Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 14 of 46Research article Cancer Biology Figure 7. MYB assembles aberrant nuclear transcription factor complexes in AML cells. (A) Volcano plot of nuclear MYB-associated proteins compared to IgG control, as analyzed by affinity purification-mass spectrometry of MV411 cells. Red symbols denote specifically MYB-associated proteins, as defined by association with CBP (MYB/IgG log2 >1). p-Values denote statistical significance of three biological replicates. (B) Enrichment of MYB- associated proteins (red) as a function of their corresponding CRISPR DepMap dependency scores in MV411 cells. Red symbols denote functionally Figure 7 continued on next page Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 15 of 46
Research article Cancer Biology
Figure 7 continued
required proteins, as defined by the genetic dependency of CBP (score < 0.18). (C) Network of BioGRID protein interactions for MYB-associated
nuclear AML proteins as a function of their respective hematopoietic expression aberrancy scores, based on their relative gene expression in AML cells
as compared to normal bone human bone marrow progenitor cells (white to red color gradient indicates increasingly aberrant gene expression). (D)
Comparison of the genetic dependency scores in leukemia cell lines as compared to all other non-hematopoietic cancer cell lines for MYB-associated
nuclear AML proteins, with red symbols denoting proteins that are required in leukemia as compared to non-hematopoietic cancers.
The online version of this article includes the following source data and figure supplement(s) for figure 7:
Source data 1. Lists of proteins identified by IP-MS of IgG control and MYB or CBP/P300 complex purifications from MV411 cell nuclei.
Figure supplement 1. CBP is the primary binding partner of MYB in AML cells.
blood cells, albeit with variable abundance, with the exception of CEBPA and SATB1 that were
measurably expressed exclusively in AML cells (Figure 9A).
To determine whether specific combinations of MYB-assembled transcription factors are associ-
ated with the leukemic activity of the MYB complex, we clustered the protein abundance values of
distinct groups of transcription factors using principal component analysis (Figure 9B–D). We found
that this approach exhibited excellent separation between CRYBMIM-sensitive (MV411, OCIAML2,
OCIAML3, HL60) and less sensitive (Kasumi1, K562) cell lines, with the first (PC1, 62%) and second
(PC2, 19%) eigenvectors explaining more than 80% of the variability in CRYBMIM susceptibility
(Figure 9B). In particular, the protein abundance levels of CEBPA and LYL1 exhibited the greatest
contribution to the observed clustering (Figure 9C), consistent with their observed activity in MYB-
assembled chromatin dynamics (Figure 6).
In addition to aberrant co-expression of various MYB-assembled cofactors, their aberrant assem-
bly in leukemia cells may also contribute to their oncogenic functions. To test this hypothesis, we
purified MYB complexes from normal human cord blood progenitor cells using immunoprecipitation,
and determined the abundance of specific cofactors as compared to human AML cells using western
immunoblotting (Figure 9E). While MYB, CBP/P300 and LYL1 were physically associated in normal
umbilical cord blood mononuclear cells, we did not observe their physical association with E2A,
SATB1 and LMO2. In contrast, MYB and CBP/P300 complexes were highly enriched in LYL1, SATB1,
E2A, and LMO2 in MV411 and HL60 AML cells (Figure 9E). Thus, oncogenic MYB transcription fac-
tor complexes are aberrantly organized in AML cells, associated at least in part with inappropriate
transcription factor expression.
If convergent assembly of MYB with common transcription factors in biologically diverse subtypes
of AML is responsible for the induction of oncogenic gene expression and blockade of normal
hematopoietic differentiation, then pharmacologic blockade of this process would suppress shared
gene expression programs associated with AML growth and survival, and promote gene expression
programs associated with hematopoietic differentiation. To test this prediction, we carried out com-
parative gene expression analyses using RNA-sequencing of CRYBMIM effects in AML1-ETO-translo-
cated Kasumi1, DNMTA3A;NPM1-mutant OCIAML3, MLL-rearranged MV411, NRAS-mutant;MYC-
amplified HL60, and CALM10-rearranged U937 AML cell lines (Figure 5A–B and Figure 9—figure
supplement 1). Unsupervised clustering of differentially expressed genes exhibited excellent separa-
tion between CRYBMIM- and PBS-treated cells. In agreement with the prediction, we observed a
shared set of genes that was suppressed in expression upon CRYBMIM treatment of all AML cell
lines, such as MYC for example (Figure 5A and Figure 9—figure supplement 1A–B). Similarly, we
observed a shared set of genes that was induced by CRYBMIM treatment, including numerous genes
associated with hematopoietic differentiation such as FOS, JUN, and ATF3, as well as SERPINE,
KLF6, DDIT3, and NFKBIZ (Figure 5A–B and Figure 9—figure supplement 1B). Thus, oncogenic
gene expression in biologically diverse subtypes of AML involves convergent and aberrant assembly
of MYB transcription factor complexes that induce genes that promote leukemogenesis and repress
genes that control cellular differentiation.
Aberrant organization of the MYB transcription factor complex is
regulated by proteolysis
We noted that MYB:CBP/P300 binding in AML cells was reduced by several orders of magnitude
upon CRYBMIM treatment (Figure 6). This contrasts to the nearly equal binding affinities of
Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 16 of 46Research article Cancer Biology Figure 8. MYB and CBP/P300 assemble convergently organized nuclear transcription factor complexes in genetically diverse AML cells. (A–B) Western blots of specific transcription factors in immunoprecipitated MYB (A) and CBP/P300 (B) nuclear complexes in seven genetically diverse AML cell lines, as indicated. Blue and red bands indicate molecular weight markers. LaminB1 serves as the loading control. Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 17 of 46
Research article Cancer Biology Figure 9. Specific MYB complex factors are aberrantly expressed and assembled in AML as compared to normal human blood cells. (A) Western blots of specific MYB complex transcription factors in normal human blood cells and genetically diverse leukemia cells, as indicated. Actin serves as the loading control. (B) Principal component analysis of MYB complex transcription factor abundance, as quantified by image densitometry, as a function of susceptibility of various AML cell lines to CRYBMIM (blue color index). (C) Contribution of individual MYB complex transcription factor abundance to the Figure 9 continued on next page Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 18 of 46
Research article Cancer Biology
Figure 9 continued
top PCA eigenvector. (D) Heatmap of hierarchical clustering of MYB complex individual transcription factor abundance and CRYBMIM susceptibility. (E)
Western blots of specific transcription factors in specific MYB nuclear complexes immunoprecipitated from normal human umbilical cord mononuclear
cells (MNC), as compared to MV411 and HL60 AML cells. Blue and red bands indicate molecular weight markers. LaminB1 serves as loading control.
The online version of this article includes the following source data and figure supplement(s) for figure 9:
Source data 1. All coding gene expression changes measured by RNA-seq in 1 hr CRYBMIM vs PBS treated 5 AML cell lines.
Figure supplement 1. MYB transcription complexes induce shared and repress distinct gene expression programs in genetically diverse AML cells, as
remodeled by peptidomimetic CRYBMIM inhibition.
CRYBMIM and native MYB to CBP/P300 KIX domain (Figure 2D), suggesting that cellular processes
must contribute to the biologic effects of CRYBMIM. Thus, we examined cellular MYB protein levels
using quantitative immunoblotting (Figure 10A, Figure 10—figure supplement 1). We found that
CRYBMIM treatment induced nearly 10-fold reductions in cellular MYB protein levels with exponen-
tial kinetics on the time-scale of 1–4 hr in CRYBMIM-sensitive MV411, HL60, OCIAML2, and
OCIAML3 cells. In contrast, CRYBMIM treatment induced less pronounced depletion of MYB in
U937, Kasumi1, and K562 cells that are less sensitive to CRYBMIM. Susceptibility of diverse AML cell
lines to CRYBMIM, as measured by cell viability (Figure 3G), was significantly correlated with the
apparent kinetics of MYB protein decay (Pearson r = 0.94; Figure 10B).
Rapid reduction of MYB protein levels by CRYBMIM is consistent with proteolysis. To investigate
this directly, we quantified CRYBMIM-induced reduction of MYB protein levels in MV411 cells upon
co-treatment with the proteosomal/protease inhibitor MG132 (Figure 10C). Consistent with the pro-
teolytic depletion of MYB upon CRYBMIM treatment, MG132 co-treatment led to near complete
rescue of this effect (Figure 10C). This effect was specific because overexpression of BCL2, which
blocks MYBMIM-induced apoptosis (Ramaswamy et al., 2018), and rescued the depletion of cellular
CREB, presumably due to non-specific proteolysis that accompanies apoptosis, was unable to rescue
CRYBMIM-induced proteolysis of MYB (Figure 10C). Since CBP is required for the anti-leukemic
effects of CRYBMIM, we queried whether MYB protein levels are affected by CBP depletion. We
observed no measurable differences in MYB protein levels between wild-type, CBP-deficient or con-
trol AAVS1-CRISPR targeted MV-411 cells (Figure 10—figure supplement 1C). This suggests that
either P300 can compensate for CBP-mediated functions, and/or MYB proteolysis is not regulated
solely by its interaction with CBP/P300, but requires specific activities of CBP/P300 induced by
CRYBMIM. In all, MYB transcription complexes are regulated by specific factor proteolysis in AML
cells and can be induced by its peptidomimetic blockade.
Release and redistribution of MYB-sequestered transcription factors
restores normal myeloid differentiation
We reasoned that the remodeling of MYB regulatory complexes and their associated chromatin fac-
tors such as CBP/P300 are responsible for the anti-leukemia effects of CRYBMIM, at least in part via
reactivation of cellular differentiation of MV411 AML cells (Figure 5). To test this, we prioritized
CEBPA, LYL1, SPI1, and RUNX1 as MYB-associated co-factors based on their physical interactions
and functional dependencies in AML cells (Figure 7), and analyzed their chromatin dynamics in
response to CRYBMIM treatment using ChIP-seq analysis. Consistent with the release and redistribu-
tion mechanism, we observed both coherent and factor-specific dynamics of MYB co-factors on
chromatin upon CRYBMIM treatment (Figure 11A). Clustering of observed dynamics revealed nine
classes of apparent chromatin responses (Figure 11B). Approximately one-third of the affected
genes lost both MYB and CBP/P300 in response to CRYBMIM, as well as RUNX1, LYL1, and/or
CEBPA (Figure 11B; yellow clusters 1, 3, and 6). Genes in the MYB and CBP/P300-depleted clusters
1, 3, and 6 were enriched in those associated with the development of hematopoietic progenitor
cells, as well as MYC and HOXA9/MEIS1 targets (Supplementary file 3c), consistent with current
and past gene expression profiling studies (Figure 5 and Figure 9—figure supplement 1). In addi-
tion, these genes were enriched in pathways involving chromatin repression, consistent with the
enrichment of DNA-binding sequence motifs of transcriptional repressors YY1 and REST/NRSF
(Figure 11B). This suggests a potential mechanism for the long-hypothesized repressive functions of
MYB. While we found no apparent changes in JUN occupancy, motif analysis at loci that both lost
Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 19 of 46Research article Cancer Biology Figure 10. Peptidomimetic remodeling of MYB transcriptional complexes leads to rapid MYB proteolysis. (A) Quantification of MYB abundance in HL60 cells as a function of duration of 20 mM CRYBMIM treatment (red) as compared to PBS control (black) using Western blot image densitometry. Lines represent single exponential decay fits. Western blots and fits for all cell lines studied are shown in Figure 10—figure supplement 1. Symbols represent biological triplicates. (B) MYB protein half-life, as estimated by exponential decay kinetics, in genetically diverse AML cell lines as a function of CRYBMIM susceptibility (Pearson r = 0.94, excluding resistant K562). Horizontal bars represent standard deviation of CRYBMIM susceptibility. Vertical bars represent standard deviation of time constants. (C) Western blots of MYB and CREB in MV411 AML cells transduced with MSCV retroviruses encoding GFP control (MV411-MSCV) or BCL2 (MV411-BCL2), treated with 20 mM CRYBMIM or PBS control with or without 10 mM of MG132 for 1 or 4 hr, as indicated. Actin serves as loading control. The online version of this article includes the following figure supplement(s) for figure 10: Figure 10 continued on next page Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 20 of 46
Research article Cancer Biology
Figure 10 continued
Figure supplement 1. CRYBMIM causes MYB proteolysis with differential degradation rates in AML cells.
and gained MYB and CBP/P300 revealed enrichment of AP-1 sequence elements, consistent with
the presence of other AP-1 family member(s) in MYB regulatory complexes. Minor apparent contri-
bution of MYB:CBP/P300-independent chromatin dynamics involved genes enriched in MLL targets
(Figure 11B, Supplementary file 3c; blue clusters 5, 8), consistent with the activity of MLL fusion
proteins in MV411 cells.
Notably, the two chromatin dynamics clusters 4 and 9 that gained both MYB and CBP/P300 in
response to CRYBMIM, associated with the accumulation of CEBPA, RUNX1, and/or CREB, were
enriched in genes controlling myeloid differentiation programs (Figure 11B; orange, and
Supplementary file 3c). This is consistent with the CRYBMIM-induced gene expression differentia-
tion programs and accompanying morphologic features of myeloid differentiation (Figure 5). While
CRYBMIM induces MYB proteolysis, residual MYB can remain bound to chromatin, as evident by its
accumulation in specific loci upon CRYBMIM treatment. MYB-binding loci lost upon CRYBMIM treat-
ment showed significant enrichment for known MYB binding motifs, while CRYBMIM-induced MYB
peaks did not (p=1e-149 vs 1e-3, 56% vs 5.7% of target sites, respectively). This raises the possibility
that DNA-binding affinity of MYB could be regulated by CBP/P300, either by direct effects such as
MYB deacetylation upon CRYBMIM treatment, or indirectly via binding with other transcription fac-
tors. It is possible that other transcription factors may contribute to oncogenic gene expression in
AML cells, such as CREB for example, as evident from their contribution to MYB-independent chro-
matin dynamics (Figure 11B; pink cluster 2). However, this is likely a minor effect, given the relatively
modest reprogramming of gene expression by CREBMIM that targets the CREB:CBP/P300 com-
plexes in AML cells (Figure 5 and Figure 5—figure supplement 1).
Globally, the most pronounced feature of MYB complex remodeling is the release of CBP/P300
from genes that are associated with AML cell growth and survival to those that are associated with
hematopoietic differentiation. To examine this, we compared relative gene expression as a function
of the relative occupancy of CBP/P300 and MYB upon 4 hr of CRYBMIM treatment (Figure 11—fig-
ure supplement 1A–B). In contrast to the model in which blockade of MYB:CBP/P300 induces loss
of gene expression and loss of transcription factor and CBP/P300 chromatin occupancy, we also
observed a large number of genes with increased expression and gain of CBP/P300 occupancy (Fig-
ure 11—figure supplement 1A–B). This includes numerous genes that control hematopoietic differ-
entiation, such as FOS, JUN, and ATF3. In the case of FOS, we observed that CRYBMIM-induced
accumulation of CBP/P300 was associated with increased binding of RUNX1, and eviction of CEBPA
and LYL1 (Figure 11—figure supplement 1C).
To confirm directly that peptidomimetic blockade of MYB:CBP/P300 releases CBP/P300 and pro-
motes its association with alternative transcription factors, we specifically immunoprecipitated MYB
and CBP/P300 from MV411 AML cells using respective antibodies, and determined their composi-
tion by western immunoblotting (Figure 12A–B). In agreement with biochemical studies, we
observed substantial depletion of CBP/P300 from immunoprecipitated MYB complexes upon CRYB-
MIM treatment (Figure 12A), and of MYB from immunoprecipitated CBP/300 complexes
(Figure 12B). Similarly, we observed reduced binding of LYL1, SATB1, E2A, LMO2, and CEBPA. In
contrast, whereas no detectable RUNX1 was found co-associated with MYB either at baseline or
upon CRYBMIM treatment (Figure 12A), assembly of RUNX1 with CBP/P300 was increased by more
than fourfold upon CRYBMIM treatment, as measured by image densitometry (p=3.4e-4;
Figure 12B–C).
In all, these results support the model in which the core regulatory circuitry of AML cells is orga-
nized aberrantly by MYB and its associated co-factors including LYL1, C/EBP family members, E2A,
SATB1 and LMO2, which co-operate in the induction and maintenance of oncogenic gene expres-
sion, as presumably co-opted by distinct oncogenes in biologically diverse subtypes of AML (Fig-
ure 13). This involves apparent sequestration of CBP/P300 from genes controlling myeloid cell
differentiation. Thus, oncogenic gene expression is associated with the assembly of aberrantly orga-
nized MYB transcriptional co-activator complexes, and their dynamic remodeling by selective
Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 21 of 46Research article Cancer Biology Figure 11. Chromatin dynamics of CRYBMIM remodeled MYB transcription factor complexes in AML cells. (A) Heatmap of transcription factor chromatin occupancy in MV411 cells as a function of time of control PBS or CRYBMIM treatment. Nine clusters identified using k-means clustering are marked by yellow (loss of MYB and CBP), purple (loss of CBP and gain of MYB), orange (gain of MYB and CBP), pink (gain of CBP), and blue (no apparent changes of MYB and CBP) boxes, based on the similarity of their z-scores, with red and blue representing enrichment or depletion of factors, Figure 11 continued on next page Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 22 of 46
Research article Cancer Biology
Figure 11 continued
respectively. (B) Groups of clusters comprising similar responses to CRYBMIM treatment based on MYB and CBP/P300 dynamics. Sequence motifs
enriched at specific loci for each cluster are listed. Factors in white denote factors presumed to be enriched based on sequence motif enrichment.
The online version of this article includes the following source data and figure supplement(s) for figure 11:
Source data 1. Differentially increased or decreased peaks measured by multiple TF ChIP-seq in 1 hr CRYBMIM vs PBS-treated MV411 cells.
Figure supplement 1. Oncogenic MYB transcription complexes sequester CBP/P300 to control AML gene expression.
blockade of protein interactions can be leveraged therapeutically to induce AML cell differentiation
and apoptosis.
Discussion
Dysregulated gene expression is a near universal feature of all human cancers. This is particularly rel-
evant for leukemias which are frequently caused by mutations of genes encoding transcription and
chromatin remodeling factors. Among all types of leukemias examined to date, the transcription fac-
tor MYB ranks as the most selectively required functional genetic dependency. This nominates MYB
both as a compelling therapeutic target, and a focus of mechanistic studies to define fundamental
mechanisms of dysregulated gene expression in leukemias.
By rapid and selective peptidomimetic interference with the binding of CBP/P300 to MYB, but
not CREB or MLL, we find that the leukemic functions of MYB are mediated by CBP/P300-mediated
co-activation of a distinct set of transcriptional factor complexes that are aberrantly assembled with
MYB in AML cells. The second-generation, cell-penetrant peptidomimetic MYB inhibitor, termed
CRYBMIM, has potent and broad-spectrum activity against diverse subtypes of AML, while relatively
sparing normal hematopoietic progenitor cells. Consequently, its improved activity enables high-res-
olution, genome-wide studies of chromatin and gene expression dynamics that control MYB-depen-
dent leukemic expression in AML cells. We find that CRYBMIM blocks oncogenic MYB gene
expression and restores myeloid cell differentiation. This effect involves aberrantly organized MYB
regulatory complexes, stably composed of additional transcription factors including LYL1, C/EBP
family members, E2A, LMO2 and SATB1, that are reminiscent of core regulatory circuits observed in
MYB-dependent T-cell lymphoblastic leukemias and other cancers (Mansour et al., 2014;
Sanda et al., 2012). Unexpectedly, we find that MYB-dependent leukemogenic gene expression
also involves apparent sequestration of CBP/P300. In turn, peptidomimetic MYB:CBP/P300 blockade
releases and redistributes CBP/P300 and other sequestered transcription factors to induce cell dif-
ferentiation. In all, these findings establish a compelling strategy for pharmacologic reprogramming
of oncogenic gene expression that supports its targeting for leukemias and other human cancers
caused by dysregulated gene control.
What is the origin of aberrant MYB transcriptional complexes and functions in leukemia cells?
MYB is not known to be mutated in most cases of AML, and this study points to its aberrant assem-
bly as the convergent mechanism by which it is pathogenically dysregulated. Indeed, previous stud-
ies have found cell-type-specific features of MYB gene activation, suggesting the presence of other
factors that influence MYB activity (Lei et al., 2004). Furthermore, MYB alone is not sufficient for leu-
kemic cell transformation, indicating the need for specific co-factors in its leukemogenic activity
(Gonda et al., 1989; Hu et al., 1991).
By integrating functional genomics and proteomics, combined with gene expression and chroma-
tin dynamics analyses, we identified a set of factors in complex with MYB that appear to be aber-
rantly and stably co-assembled, including C/EBP family members, LYL1, E2A, LMO2, and SATB1.
Their physical interactions and chromatin co-localization with MYB are associated with oncogenic
gene expression and blockade of cell differentiation in AML cells. Interestingly, we found no CRYB-
MIM-induced remodeling of MYB regulatory complexes independent of CBP/P300, indicating that
KIX-dependent interaction between MYB and CBP/P300 is required for most of MYB transcriptional
activity in AML cells.
It is possible that somatic mutations of regulatory DNA elements, such as those physically associ-
ated with MYB regulatory complexes, contribute to the aberrant assembly of these complexes on
chromatin, as observed for the oncogenic TAL1 enhancer mutations in cases of T-ALL
Takao, Forbes, Uni, et al. eLife 2021;10:e65905. DOI: https://doi.org/10.7554/eLife.65905 23 of 46You can also read
